Jonathan Wells' weird notions about development

Jonathan Wells recently gave a talk in Albuquerque at something called the "Forum on Science, Origins, and Design", a conference about which I can find absolutely nothing on the web. I wasn't there, of course, and I don't get invited to these goofy events anyway, but I did get a copy of Wells' powerpoint presentation from an attendee. It's titled "DNA Does Not Control Embryo Development" — shall we look at it together? It's really a hoot.

Jonathan Wells has been exposed to a little bit of knowledge, just enough for him to regurgitate a few common phrases that are current in the developmental biology literature, and to pretend that he is profound and revolutionary. Nothing could be further from the truth — his title comes straight out of mainstream evo-devo. It's actually rather funny, since there is a tiny bit of truth in what he says, but what's true isn't original with him, and it certainly doesn't have the implications he thinks.

He starts with a few quotes to show that some scientists tend to have both a naive view of development, and tend to simplify a bit when speaking in public. Here's James Watson:

We know that the instructions for how the egg develops into an adult are written in the linear sequence of bases along the DNA of the germ cells.

And Ken Miller:

Our genome is an "amazing script, carrying within it all of the instructions for building a human being."

And Francis Collins:

The recipe for building the animal body is controlled by remarkably few genes.

Woo-hoo. These are all good scientists who are _not_ developmental biologists, and who are guilty of the dreadful sin of over-simplification. It's entirely true, however, that there is a tendency to reduce the role of the genes to an excessively narrow view of "control", and it's an attitude that many modern biologists would like to see broadened.

Wells missed a trick, though. There's an even better source for this idea, in Richard Lewontin's excellent little book, _The Triple Helix: Gene, Organism, and Development_^{([amzn](http://amazon.com/exec/obidos/ASIN/0674006771/pharyngula-20)/[b&n](http://service.bfast.com/bfast/click?bfmid=2181&sourceid=41505966&bfpid=0674006771&bfmtype=book)/[abe](http://www.dpbolvw.net/click-1772498-9836638?url=http%3A%2F%2Fwww.abebooks.com%2Fservlet%2FSearchResults%3Fisbn=0674006771&cm_ven=CJ&cm_cat=1616003&cm_pla=1772498&cm_ite=Abebooks-Book+Redirection+Allowed)/[pwll](http://www.powells.com/cgi-bin/partner?partner_id=30010&cgi=product&isbn=0674006771))}:

One of the most eminent molecular biologists, Sydney Brenner, speaking before a group of colleagues, claimed that if he had the complete sequence of DNA of an organism and a large enough computer then he could compute the organism. The symbolic irony of this remark is that it was made in the opening address of a meeting commemorating the 100th anniversary of Darwin's death. A similar spirit motivates the claim by yet another major figure in molecular biology, Walter Gilbert, that when we have the complete sequence of the human genome "we will know what it means to be human." Just as the metaphor of development implies a rigid internal predetermination of the organism by its genes, so the language used to describe the biochemistry of the genes themselves implies an internal self-sufficiency of DNA.

The rest of the book is very good, and continues along the same lines to point out the interdependency of genes, cytoplasm, and environment.

Oh, but wait — Wells can't use that. If his point is that he and his creationist ilk are clever iconoclasts, it does no good to reveal that there are huge numbers of biologists, including evolutionary biologists, who have been arguing that the metaphor of control is invalid.

Besides, it's silly for an IDist to be complaining about metaphor. The primary foundation of their rationalizations is an extended and inappropriate metaphor, that you can call molecules "machines" and pretend that that means they were built by designers in little factories.

Wells has a grander plan than to rebuke a few molecular biologists for sloppy language, though. Here's the key point in his talk.

See the "logic"? If some people think DNA controls development, and if those same people accept evolution, than by showing that DNA does not control development, then evolution must be false. It's Wells' familiar game of throwing dirt in random directions and hoping it will stick on a few biologists, so he can claim the whole field is wrong. It's a rather absurd premise.

It'll be interesting to see how he plans to claim this erroneous metaphor is central to "Darwinism", since Darwin did not know anything DNA or genes, and didn't have this metaphor at all. One solution, of course, is to just skip over the first 60 or 70 years of evolutionary biology and jump to the neo-Darwinian synthesis, when genes were incorporated into the theory.

Uh, OK. It's a summary of the synthesis that doesn't mention Fisher and Wright and population genetics, Dobzhansky on genetics and speciation, Mayr and biogeography, Simpson and paleontology. It's a bit cartoonish, especially since the first two points are purely Darwinian and precede the synthesis, and the fourth was also worked out prior to the synthesis by people like Morgan and Muller. And the third point…hey, wait a minute. Where'd that come from? That isn't part of the neo-Darwinian synthesis!

Genes (DNA) carry all the essential hereditary information, and a "genetic program" controls embryo development.

Wells just tossed that in. We know that is not true, and we've known it for a long, long time. I mean, Wells is supposed to be a developmental biologist — he must know about, to name one example, the work of EG Conklin on ascidians around 1900, in which he mapped out the distribution of cytoplasmic determinants in the egg. The synthesis said almost _nothing_ about development, and here's Wells trying to pull a fast one and tell you that not only was development a key underpinning of the theory, but that a false metaphor was specifically part of the story.

It wasn't. He's making stuff up.

Shall we see some more made-up stuff?

Heh. First day of class, first lecture in my developmental biology course, all I talk about is [genomic equivalence](http://www.ncbi.nlm.nih.gov/books/bv.fcgi?rid=dbio.section.547). It's very basic stuff: the cells in your body all have the same genes, but different cellular phenotypes. That's the cool stuff in development, thinking about the way cell fates switch under regulatory control.

I don't call it a "paradox", though. Why would I? It's like saying that the existence of "if" statements in computer science is a paradox! Conditional expression of subsets of genes is central to development — it's hard to imagine differentiation occurring without it.

Here's another weird one.

This is a complete misrepresentation of the Klotz paper — it says no such thing. Here's what the paper is actually about.

In addition to the DNA in the nucleus, cells also need many other components to function. One set of these extranuclear elements are various proteins associated with the cytoskeleton, an internal framework of tubular rods that run throughout the cell. In particular, there is a specific structure called the centrosome that is a kind of cytoskeletal organizing center — sort of like a seed crystal that acts as a starting point for growth. One special role for the centrosome is as an anchor point for the fibers upon which chromosomes shuttle about during cell division.

Sperm cells carry a centrosome as well as a nucleus. That centrosome appears to be important in triggering cell divisions. What Klotz and his coinvestigators did was take unfertilized frog eggs, prick them to trick them into responding as if sperm entry had occurred, and injected them with the centrosomic proteins. They demonstrated that the centrosomes were important for driving early cell divisions — they did not get complete frogs. Cell divisions eventually arrested, but some of the pseudo-fertilized eggs got as far as the tadpole stage.

We do similar things with zebrafish. Sometimes you want embryos that have only the maternal genes, and the way you do this is to suppress a meiotic division (so that the egg will be diploid rather than haploid), and fertilize it with sperm that has been so thoroughly irradiated that its genetic material has been destroyed, and all it is doing is carrying a centrosome and some other activating factors into the egg.

These experiments emphatically do not demonstrate that DNA does not matter — the cells have a full complement of maternal DNA. It does not demonstrate that the paternal centrosome is sufficient to generate a complete animal. This claim is complete bunk. Don't just take my word for it: here's the conclusion to the Klotz paper.

Taken together, these results are consistent with the idea that the whole or part of the centrosome structure acts as a seed to start the centrosome duplication cycle in _Xenopus_ eggs.

That's a more modest and far more appropriate interpretation of the results.

Next!

Cool. Hox genes are always fun. Wells then gives an overview of the classification of genes by Nusslein-Volhard and Wieschaus, who found that there were early maternal effect genes, gap genes that block out broad chunks of the anterior-posterior axis, segmentation genes that partition the embryo into repeating elements, segment polarity genes that assign identity within a segment, and then the Hox genes, which give the segments special identities in the body plan. And then, this is his conclusion:

This is an old theme for Wells. Hox genes aren't the very first genes turned on in the embryo, so they must not be that important; in his previous books, he's tried to make a similar case that because Haeckel's phylotypic stage (the pharyngula in chordates) isn't the earliest stage, and that there is more variation between classes of organisms in the earlier stages, evolution is false. It doesn't follow.

Hox genes are activated _relatively_ late in _early_ development. That's still early, at 4-5 weeks of development in humans. There are many other genes that are expressed later still, and they are seriously dependent on an appropriate earlier pattern of Hox gene expression. Developmental processes build on prior processes; this isn't surprising at all.

It is true that flies have some positional information laid out before fertilization. This is also not a new revelation; I already mentioned the work of Conklin on ascidians, so we've known for over a century that many animals are mosaic, getting a headstart on spatial differentiation with a specific distribution of some informational molecules. This isn't magic (AKA intelligent design), however — it's generated by an asymmetrical arrangement of supporting cells in the ovary. Not all animals do this. One example: mammals! Our eggs don't seem to have much, if any, spatial information initially, and all of the cells in the early blastula are mostly equivalent. Differences emerge progressively by interactions with each other and with neighboring tissues.

This next slide is amusing.

Next time I see Fred Nijhout, I'm going to have to let him know that he is being cited by creationists. I think he'll find it funny. Maybe.

That quote is from an excellent paper by Nijhout in which he takes the abuse of metaphors to task, and points out the flaws of the "control" and "program" models of the genome. Here's the abstract.

In describing the flawless regularity of developmental processes and the correlation between changes at certain genetic loci and changes in morphology, biologists frequently employ two metaphors: that genes 'control' development, and that genomes embody 'programs' for development. Although these metaphors have an admirable sharpness and punch, they lead, when taken literally, to highly distorted pictures of developmental processes. A more balanced, and useful, view of the role of genes in development is that they act as suppliers of the material needs of development and, in some instances, as context-dependent catalysts of cellular changes, rather than as 'controllers' of developmental progress and direction. The consequences of adopting this alternative view of development are discussed.

Re-reading this paper, it looks like many of the correct ideas in Wells' talk are lifted directly from it; the incorrect ones seem to be Wells' own weird biased misinterpretation. What's important, though, is that it actually invalidates Wells' overall conclusion. He wants to claim that evolution is built partially on this premise of genetic "control", yet here is a paper by a well-known evolutionary developmental biologist not only explaining that "control" is a poor metaphor, but discussing how evolutionary theory is made richer by a better and more accurate understanding of developmental processes! It's as if Wells is ignoring everything Nijhout was writing _except_ for a few statements that he could lift out of context and pretend that they are an attack on the foundation of evolutionary thought. (Yeah, I know. Big surprise.)

Let's deal with Wells' conclusion, finally.

**The floor plan of the embryo is in the form of spatial information that precedes DNA differentiation.** This is not always true. It's only the case in some animals. It doesn't matter, though, since maternal factors are also the product of evolution…and are also produced by maternal genes.

**Some of this spatial information is carried by the centrosome and cortex.** This is true. It's been recognized by biologists for a long, long time. Once again, the ID research program uncovers the obvious that real biologists have been discussing for a century.

**There is evidence that both the centrosome and cortex are heritable independently of the DNA.** Well, sort of true. Also sort of wrong. He's making the complement of the mistake he's accusing biologists of making. Just as you can't say all the hereditary information is confined to DNA, you also can't say that it's the domain of the centrosome or cortex. The centrosomal proteins are _gene products_! What we're really dealing with is a tangled web of interdependent processes.

**So DNA does not determine all the essential characteristics of living things. It is necessary but not sufficient for embryo development.** Yes. Again with the obvious, Mr Wells! Look, vehicle. Look, road. Which one is in charge of the route you will take from point A to point B? You need both. And biologists, including evolutionary biologists, have been aware of this for a long time.

**The idea that genetic programs control embryo development is not an inference from evidence, but a deduction from neo-Darwinian evolutionary theory.** Breathtaking nonsense. Neo-Darwinian evolutionary theory says nothing of the kind, especially since it says almost nothing about development at all. You can actually see classical evolutionary biology as only saying there is a measurable correlation between the hereditary material (whatever it is; it wouldn't even have to be nuclear) and the phenotype. The causal mechanisms that translate genes into form and function is the domain of developmental biology, which most other biological disciplines have simply treated as an abstraction.

I'd like to know what Wells actually said at his conclusion. I have to infer from his slides that he's making a strange argument that heredity is not the product of DNA, and therefore the evolutionary evidence from genes and molecules is invalid. I know, that sounds awesomely stupid and it's hard to believe anyone could stand up in public and say it, but that's the thrust of his presentation. Then again, Wells has been willing to say lots of stupid things in his books, so I shouldn't be surprised that he'd do likewise in his lectures.